Robots Took Your Jobs, Not Foreigners

The experts are right. The pundits? Not so much…
Assembly line robots weld the front cab of Chrysler’s 2009 Dodge Ram pickup in Warren, Michigan — Dec. 17, 2008 (AP Photo/Carlos Osorio, file)

Assembly line robots weld the front cab of Chrysler’s 2009 Dodge Ram pickup in Warren, Michigan — Dec. 17, 2008 (AP Photo/Carlos Osorio, file)

People love to declare something to be common sense in debates. It’s our go to shorthand for saying something is obvious on its face. But when a pundit delivers the kind of folksy, down home wisdom that older viewers of cable news applaud, is that supposed common sense actually right? We tend to gravitate to what sounds agreeable to us, things that play into our views, giving our brain a break from constantly restructuring how we perceive the world, and the brain rewards us for allowing it to be lazy. Unfortunately for our slothful minds, the world runs on facts and neglecting this in favor of a soothing status quo isn’t “common sense” or “home-spun wisdom” as cable news would have us believe. It’s complacency. And entire empires have been brought down by refusing to tackle real crises by partying as if their fortunes would never change, or fighting imaginary problems while the real ones ate away at their institutions and infrastructures, physical and political.

When British politician Michael Gove declared that the public had enough expert opinions in his defense of the Brexit, it was a given that millions of people who voted for him and his allies agreed because to them, he wasn’t just discarding facts and data, but urging people to revert to that cozy, lazy, homey common sense. For people whose future prospects were savaged by changes and agreements that sound like technobabble to them, it can be an infuriating experience to watch an expert in a suit pontificate in jargon, and doubly enraging when he or she shrugs and says that some people just lack the skills to keep up with where the world is going. Meanwhile, many of the communities they know hardly changed beyond more people moving out to take their chances elsewhere, and landmarks slowly falling into disrepair. In their minds, the world isn’t different, greedy elites and smarmy pundits are just talking nonsense on the tube to justify their shenanigans.

But is that true? In a word, no. We can see the chasm between a common sense explanation for vanishing manufacturing and entry level jobs, and an expert approach that looks at long term data and future trends. The news-ready refrain is that outsourcing and bringing in immigrants on H1-B visas has been killing jobs left and right. Unfair competition from nations where worker safety rules are lax to nonexistent, and currencies are gamed to stay too low and are taking away manufacturing, say pundits as they bring up dire graphs on screen. As an additional follow-up to their horror stories, they’ll talk about the real abuses and fraud that happen with workers brought in with H1-Bs while pushing out older native employees in dignity-stripping mass layoffs. However, while outsourcing and visa abuses are definitely making a big dent, they pale in comparison to the impact of ever more sophisticated software and robotics.

Not only are nearly 9 in 10 manufacturing job losses due to automation, but the very people blamed for taking away those jobs are in line to suffer the same fate. Likewise, in the U.S., H1-B visas have annual caps of 85,000, which is a little more than a rounding error when estimating the American workforce. What that means is if we were to wave a magic wand and end all outsourcing and skilled immigration programs, close to 90% of people who lost their jobs wouldn’t get them back. How can they? Those jobs no longer exist and pretending that they do is a huge disservice to the public. But this doesn’t play well on the campaign trail, so populist avatars rail against evil outsourcers and companies importing cheaper labor from overseas instead of trying to figure out how to integrate those who are out of work into the new, globalized, mechanized labor force. Even worse, the very people who would need to be in charge of these efforts will blatantly deny the facts to the experts’ astonishment, effectively lying to their desperate base.

The sad truth is that any job involving traditional paper pushing, document reviews, or routine tasks are frighteningly easy to automate today. Reams of paperwork once processed by armies of office paper pushers are now trivial to replace with slick, easy to use websites which prevent errors that could validate the submitted data, collect payments, and pass it on to software that can review the information and render decisions in milliseconds. Even the rapidly decreasing amount of forms filled out by hand can be digitized entirely by computers that use convolutional neural networks to understand handwriting. Using this algorithm, Google knows what image you’re trying to find and your ATM understands what’s written on the check you placed in it, and if you endorsed it properly. Even research and data modeling usually done by interns and middle managers can be tackled by AI which can find a desired pattern in seconds after being fed enough training material.

Common sense would tell you that computers have limitations and that you always need humans to double check the computer’s work. But expertise in computer science will tell you where those limits are and that the tasks we farmed out to machines are being done far more accurately than any human could ever do because computers don’t have momentary lapses in memory, or concentration, they’re never tired, hungry, or stressed out, and once you dial in the right logic and workflow with code, humans just need to get out of the way until a rare error does occur, usually caused by a human entering the wrong information somewhere down the line. And based on knowing a lot about both the potential and the current state of our computers, experts can even tell you the odds are of your job being automated in the very near future. In case you’re wondering, 45% of all jobs are at risk of being done by robots or code by 2035, which isn’t exactly all that far in the future.

Non-experts don’t know exactly what computers can do and AI can sound a bit like dark magic, while jargon like convolutional neural network or naive Bayes classifier seem like pure mumbo-jumbo. It’s not their job to track the incremental and revolutionary advances in computing and its applications, or study and implement how companies and governments can slowly use a new machine or piece of software to speed up their workflows while two or three employees at a time lose their jobs to attrition, then layoffs. And they won’t even see a computer in someone’s place because the software will live on a server, somewhere in one of those cold back rooms filled with all those humming boxes on racks draped in plastic wire. What they do see is people in suits on TV talking about how they “streamlined their operations at very significant cost savings” by outsourcing to India, China, or Vietnam, and the boss telling them that foreign workers are simply more adaptable than them before firing them, saying they need to “right-size the workforce.”

So when some expert, also in a suit and tie, comes on TV or is quoted in a media outlet talking about how jobs are being lost primarily to automation, they fight to urge to scream “No they’re not, you out of touch fucking idiots in your ivory towers! Come to my office and see people being replaced by a guy who obviously lied on his resume or have their jobs sent off to China!” because that’s the tip of the iceberg they see bobbing above water. But the fact that outsourcing and often abused immigration visas impact plenty of jobs in a dramatic and visible fashion don’t make the experts wrong. They aren’t saying not to worry about outsourcing or being replaced by a cheap foreign worker if you’re in a position vulnerable to that, they’re saying “hey, if you think that’s bad, check out what’s even worse for jobs!” and they’re saying it because they studied the issues and understand that if we use our limited resources in fighting 12% of the problem, the other 88% will make mincemeat out of us while we’re busy patting ourselves on the back.

Experts are not the enemy. They have warned us about automation and its future reach for years now, but there’s only so much they can do in the role of advisers and strategists. They don’t set budgets, pass laws to reform key systems, or issue executive orders. If politicians easily swayed by the folksy pundits that infest news shows and sell their unique yet disturbingly similar brands of “common sense” won’t make the changes recommended by experts, or flatly deny a problem about which they’ve been sounding alarm after alarm, people will be worse off at the end of the day. When you hear another politician say “we had enough of experts” what they’re really saying is “I’m just trying to please people who are mad at me so they don’t vote me out of office next election,” and when talking heads on the news start yammering about how common sense tells us the experts are wrong, what they’re really saying is “let me soothe you for higher ratings.”

Today, the media is quick to treat people with actual skills, knowledge, and experience as pariahs, and uncomfortably large groups of people cheer that the eggheads and know-it-alls are “being put back in their place” while the folksy, simplistic broadsides on large, complex, festering problems that were often worsened by years of neglect or incompetent, tone-deaf handling that ignored expert advice, are presented as obviously correct. Then, when these common sense approaches fail or backfire, the politicians and talking heads all shrug and say that now they suddenly “realized it’s not so easy.” And do you know who would’ve told them it’s not and what might be a better and probably more successful approach, as well as filled them in on the history of the problem and why it became a problem in the first place? All of those “out-of-touch-with-the-common-man” experts they love to deride.

Politech // Artificial Intelligence / Machine Learning / Politcs / Tech